SAWL - Semantics-aware Location cloaking

The SAWL system

SAWL supports the evaluation and comparison of semantic location algorithms (see below) on both simulated and real spaces. The main components are:
  • Space simulator: generation of the probability grid based on a model inspired by the Community-based Mobility Model (Musolesi&Muscolo 2007). The space is then populated by either imported or randomly generated sensitive places.
  • Cloaking engine: provides an extensible set of semantic location cloaking method.

System architecture

Scenario: protecting sensitive semantic locations in LBS

  • Users visit semantic locations (i.e.,geographical places) of different type (e.g., entertainment places)
  • Semantic locations { f1,..fn} have different popularity: P(x in f1),...,P(x in fn)
  • Semantic locations can be sensitive
  • Privacy thread: associating users with sensitive semantic locations (e.g., hospitals) with a probability higher than a given threshold

Computational model and cloaking techniques

Goal: to bound the probability of user's association with sensitive locations while preserving QoS

Approach:

  1. separating the generation of cloaked regions from the user's position enforcement in order to prevent reverse engineering attacks;
  2. personalizing privacy requirements through the privacy profile;
  3. utility measures;
  4. generating cloaked regions by expanding sensitive seeds